Search

In order to provide wireless services for the current demand of high data rate mobile applications, more spectrally efficient systems are needed. As a matter of fact, the current wireless systems are limited by a frequency splitting spectrum management which in one hand minimizes the multiuser interference but; on the other hand, it precludes the use of wider bandwidth signals. As a more aggressive frequency reuse is targeted (ideally, all transmitters might eventually share the same frequency band), the use of multiple antennas for interference reliving, jointly with a smart power allocation is compulsory. In addition, novel spectrum management regulatory policies are required for ensuring a peaceful coexistence between adjacent spectrum sharing networks and for promoting their development. The aim of this dissertation is provide a beamforming and power allocation design for these novel spectrum sharing systems which are meant to exponentially increase the spectral efficiency of the systems. A mathematical framework based on multicriteria optimization for analyzing the beamforming design is provided which serves as a fundamental tool for describing the state-of-the-art studies in multiantenna interference networks. Indeed, the achievable rates are described and several ways of computing the Pareto rate region of MISO interference channel (i.e. the communication model that represents the spectrum sharing network when the transmitters use multiple antennas) are studied. Nevertheless, as the system designer aims to work in a single efficient rate point, the sum-rate optimal beamforming design is studied. Curiously, it results that under some realistic assumptions on both the desired and interference power levels, the obtained beamformer is the reciprocal version of a known receiving one and it optimizes a notion of antenna directivity for multiuser communications. Neverthelss, it is important to remark that the higher transmit power is used, the more interference dominated is the medium, not only within the wireless network, but also to eventually adjacent networks that might suffer from inter-network interference. In order to cope with this problem, a spectrum licensing system is revisited, namely time-area-spectrum license. Under this spectrum management mechanism, a license holder is able to radiate signals under a certain portion of time, within a concrete area and in a given band. Moreover, the amount of signal strength within the area is constraint by a certain value. Since controlling the signal power levels in a given area is cumbersome, we propose to restrict the receive power as an estimation of the overall accumulated signal strength. Therefore, the optimal transmit beamformers and power allocations are studied. Concretely, the achievable rates are derived and an operational working point is envisaged. In addition, a suboptimal yet low computationally complex and decentralized beamforming design is presented and it shows a good performance in front of other decentralized designs.

[ANGLÈS] This thesis deals with the problem of finding optimal precoding techniques that, considering multiple antennas at the transmitter side, will be able to increase the data rate of future interference networks. First, we present a transmit beamformer that obtains higher sum-rates of the current decentralized designs in asymmetric channel link gains yet preserving a good performance in the general case. From a more theoretical perspective, we derive the achiveable rates of the multiple-input-single-output interference channel via the mathematical framework of multicriteria optimization. Although few results have appeared for solving this problem, we show that all of them can be reformulated as a multicriteria optimization problem and we show their equivalence. Finally, we consider the problem of confidential communication in interference networks that might be required by several wireless communication applications. Relaying on physical layer formulation of the security, we obtain the weighted region when the nonintended receivers do not collude. This is, once each eavesdropper has a weighting factor, we derive the maximum rates that the transmitter can securely communicate with the intended receiver. [CASTELLÀ] Esta tesis versa sobre el problema de encontrar las técnicas óptimas de precodificación que, teniendo en cuenta múltiples antenas en el lado del transmisor, será capaz de aumentar la velocidad de datos de redes de injerencia en el futuro. En primer lugar, se presenta un formador de haces de transmisión que obtiene mayor suma de las tasas de los diseños actuales descentralizados de enlace de canal asimétrica, sin embargo, las ganancias preservan una buena actuación en el caso general. Desde una perspectiva más teórica, se derivan las tasas de achiveable de la interferencia de canales múltiples de entrada y salida única a través de la estructura matemática de la optimización multicriterio. Aunque pocos resultados han aparecido para resolver este problema, nos muestran que todos ellos pueden ser reformulados como un problema de optimización multicriterio y mostramos su equivalencia. Por último, consideramos el problema de la comunicación confidencial en redes de injerencia que podrían ser necesarias por varias aplicaciones de comunicación inalámbrica. La retransmisión en la formulación de la capa física de la seguridad, se obtiene la región ponderada cuando los receptores nonintended no concuerdan. Esto es, una vez al espía tiene un factor de ponderación, se derivan las tasas máximas que el transmisor de forma segura se pueden comunicar con el receptor previsto. [CATALÀ] Aquesta tesi tracta el problema de trobar les tècniques òptimes de precodificació que, tenint en compte múltiples antenes en el costat del transmissor, serà capaç d'augmentar la velocitat de dades de xarxes d'ingerència en el futur. En primer lloc, es presenta un formador de feixos de transmissió que obté una major suma de les taxes dels dissenys actuals descentralitzats d'enllaç de canal asimètrica, però, els guanys preserven una bona actuació en el cas general. Des d'una perspectiva més teòrica, es deriven les taxes d'achiveable de la interferència de canals múltiples d'entrada i sortida única a través de l'estructura matemàtica de l'optimització multicriteri. Encara que pocs resultats han aparegut per resoldre aquest problema, ens mostren que tots ells poden ser reformulats com un problema d'optimització multicriteri i mostrem la seva equivalència. Finalment, considerem el problema de la comunicació confidencial en xarxes d'ingerència que podrien ser necessàries per diverses aplicacions de comunicació sense fils. La retransmissió en la formulació de la capa física de la seguretat, s'obté la regió ponderada quan els receptors nonintended no concorden. Això és, un cop l'espia té un factor de ponderació, es deriven les taxes màximes que el transmissor de forma segura es poden comunicar amb el receptor previst.

An aggressive frequency reuse is expected within the next years in order to increase the spectral ef¿ciency. Multiuser interference by all in-band transmitters can create a communication bottleneck and, therefore, it is compulsory to control it by means of radiated power regulations. In this work we consider received power as the main way to properly measure radiated power, serving at the same time as a spectrum sharing mechanism. Taking into account the constraints on the maximum total receive-power and maximum transmit-power, we ¿rst obtain the transmit powers that attain the Pareto-ef¿cient rates in an uncoordinated network. Among these rates, we identify the maximum sum-rate point for noise-limited scenarios. Next, in order to reach this working point using as less power as possible, we design a novel beamformer under some practical considerations. This beamformer can be calculated in a non-iterative and distributed fashion (i.e. transmitters do not need to exchange information). We evaluateour designby meansof Monte Carlosimulations, compare it with other non-iterative transmit beam formers and show its superior performance when the spectrum sharing receive-power constraints are imposed.